run_ensemble_ler() crashes on remote distributed cluster

[tekknosol]

Hi,
I am working on a remote cluster with multiple nodes created by future::plan(cluster). Running LakeEnsemblR 1.0.5 in this environment crashes when calling the model executables.
I figured this is due to the way do.call() is used in the function run_ensemble_ler(). According to the documentation of the future package its advised to specify the function as the object itself and not by name when using do:call().

So instead of the current code used in run_ensemble.R...

model_out <- setNames(
      lapply(model, function(mod_name) do.call(paste0(".run_", mod_name),
                                               run_model_args)),
      model
    )

... this problem can be solved by doing something like:

model_out <- setNames(lapply(model, function(mod_name){
      if(mod_name == "Simstrat"){
        do.call(.run_Simstrat, run_model_args)
      }else{if(mod_name == "GLM"){
                 do.call(.run_GLM, run_model_args)
              }}
    }), model)

Probably this if() cascade is not the most elegant solution and something like e.g. a switch() statement would be better.
Anyway, I am happy to submit a pull request if you provide some thoughts on how you would like to deal with it!

[tadhg-moore]

Hey, that is a good catch as we have not yet ran LakeEnsemblR in a cluster environment and that looks like a relatively straightforward adjustment, we would be happy to have a PR with a fix on that. Apologies for the delay in response.

[tekknosol]

I realized meanwhile that some additional issues have to be solved to run LER on the cluster (at least in my case). I had to deactivate some parts of the code and I can only use the csv output mode because there seems to be a problem with the upstream NetCDF library.
But I am successfully using LER on the cluster now...

I'll try to figure it out more systematically and then open respective issues/PRs